How to maintain context with OpenAI API without refeeding?

How can I create a chatbot capable of addressing specific inquiries about product data while maintaining context and offering consecutive responses, similar to chat.openai.com, without relying on refeeding previous responses to avoid token limits?